Classification system training
First Claim
1. A non-transitory computer-readable medium having stored thereon computer-readable instructions that when executed by a computing device cause the computing device to:
- compute a baseline penalty value using a plurality of observation vectors, wherein each observation vector of the plurality of observation vectors includes an explanatory variable value and a response variable value, wherein the baseline penalty value is inversely proportional to a square of a maximum explanatory variable value;
compute a set of penalty values based on the computed baseline penalty value;
for each penalty value of the set of penalty values,train a classification type model using the respective penalty value and the plurality of observation vectors to compute parameters that define a trained model, wherein the classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector;
validate the trained classification type model using the respective penalty value and the plurality of observation vectors to compute a validation criterion value for the trained classification type model that quantifies a validation error; and
store the computed validation criterion value, the respective penalty value, and the computed parameters that define a trained model to the computer-readable medium;
determine a best classification model based on the stored, computed validation criterion value of each trained classification type model; and
output the respective penalty value and the computed parameters associated with the determined best classification model for predicting a new response variable value from a new observation vector.
1 Assignment
0 Petitions
Accused Products
Abstract
A computing device trains models for streaming classification. A baseline penalty value is computed that is inversely proportional to a square of a maximum explanatory variable value. A set of penalty values is computed based on the baseline penalty value. For each penalty value of the set of penalty values, a classification type model is trained using the respective penalty value and the observation vectors to compute parameters that define a trained model, the classification type model is validated using the respective penalty value and the observation vectors to compute a validation criterion value that quantifies a validation error, and the validation criterion value, the respective penalty value, and the parameters that define a trained model are stored to the computer-readable medium. The classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector.
19 Citations
30 Claims
-
1. A non-transitory computer-readable medium having stored thereon computer-readable instructions that when executed by a computing device cause the computing device to:
-
compute a baseline penalty value using a plurality of observation vectors, wherein each observation vector of the plurality of observation vectors includes an explanatory variable value and a response variable value, wherein the baseline penalty value is inversely proportional to a square of a maximum explanatory variable value; compute a set of penalty values based on the computed baseline penalty value; for each penalty value of the set of penalty values, train a classification type model using the respective penalty value and the plurality of observation vectors to compute parameters that define a trained model, wherein the classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector; validate the trained classification type model using the respective penalty value and the plurality of observation vectors to compute a validation criterion value for the trained classification type model that quantifies a validation error; and store the computed validation criterion value, the respective penalty value, and the computed parameters that define a trained model to the computer-readable medium; determine a best classification model based on the stored, computed validation criterion value of each trained classification type model; and output the respective penalty value and the computed parameters associated with the determined best classification model for predicting a new response variable value from a new observation vector. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. A computing device comprising:
-
a processor; and a non-transitory computer-readable medium operably coupled to the processor, the computer-readable medium having computer-readable instructions stored thereon that, when executed by the processor, cause the computing device to compute a baseline penalty value using a plurality of observation vectors, wherein each observation vector of the plurality of observation vectors includes an explanatory variable value and a response variable value, wherein the baseline penalty value is inversely proportional to a square of a maximum explanatory variable value; compute a set of penalty values based on the computed baseline penalty value; for each penalty value of the set of penalty values, train a classification type model using the respective penalty value and the plurality of observation vectors to compute parameters that define a trained model, wherein the classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector; validate the trained classification type model using the respective penalty value and the plurality of observation vectors to compute a validation criterion value for the trained classification type model that quantifies a validation error; and store the computed validation criterion value, the respective penalty value, and the computed parameters that define a trained model to the computer-readable medium; determine a best classification model based on the stored, computed validation criterion value of each trained classification type model; and output the respective penalty value and the computed parameters associated with the determined best classification model for predicting a new response variable value from a new observation vector. - View Dependent Claims (16)
-
-
17. A method of providing training of classification models, the method comprising:
-
computing, by a computing device, a baseline penalty value using a plurality of observation vectors, wherein each observation vector of the plurality of observation vectors includes an explanatory variable value and a response variable value, wherein the baseline penalty value is inversely proportional to a square of a maximum explanatory variable value; computing, by the computing device, a set of penalty values based on the computed baseline penalty value; for each penalty value of the set of penalty values, training, by the computing device, a classification type model using the respective penalty value and the plurality of observation vectors to compute parameters that define a trained model, wherein the classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector; validating, by the computing device, the trained classification type model using the respective penalty value and the plurality of observation vectors to compute a validation criterion value for the trained classification type model that quantifies a validation error; and storing, by the computing device, the computed validation criterion value, the respective penalty value, and the computed parameters that define a trained model to the computer-readable medium; determining, by the computing device, a best classification model based on the stored, computed validation criterion value of each trained classification type model; and outputting, by the computing device, the respective penalty value and the computed parameters associated with the determined best classification model for predicting a new response variable value from a new observation vector. - View Dependent Claims (18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30)
-
Specification